AI-Native Autonomous Electronic Warfare — L3Harris and Shield AI Complete First Live DiSCO Flight Test

As we wrote last week, machine learning is transforming how militaries control the electromagnetic spectrum. The theory just passed a live fire test: L3Harris and Shield AI completed the first flight of an autonomous electronic warfare ecosystem where AI, not a human operator, detected electromagnetic threats and rerouted drones in real time.

The electromagnetic spectrum has become the central terrain of modern conflict. Every radar, communication link, GPS signal, and drone datalink depends on uncontested spectrum access. The side that reads the spectrum faster wins. Until now, electronic warfare crews still sat in the loop, interpreting threat data and issuing commands. The July 16 flight test removed them.

What the Flight Test Demonstrated

The test used L3Harris' Green Wolf unmanned aircraft as the primary platform, equipped with three integrated systems: the DiSCO electromagnetic battle management ecosystem, Shield AI's Hivemind mission-autonomy software, and the Deceptor software-defined electronic warfare payload. Multiple UAVs carrying Deceptor payloads operated in a contested electromagnetic environment. The system fused threat intelligence from the distributed sensor network into a common operating picture of the spectrum. Hivemind, running onboard the Green Wolf, evaluated the threat data and autonomously commanded the aircraft and follow-on UAVs to reroute through a safe operating zone, all without human input.

The test followed a hardware-in-the-loop simulation in February 2026. Moving from simulation to live flight in five months is unusually fast for defense aerospace, where development cycles typically span years. "This successful demonstration shows how quickly we can transform concepts into operational capability for the joint force," said Lauren Barnes, President of Spectrum Superiority at L3Harris.

Why AI Autonomy Matters in Electronic Warfare

Electronic warfare moves at the speed of light. A human operator in a ground station, even with low latency, adds seconds to the detect-decide-act cycle. In spectrum operations, seconds are enough for an adversary to complete an electronic attack or shift frequencies. Hivemind compressed that cycle to real-time machine speed, enabling the unmanned systems to sense, share, and act on spectrum threats faster than any human-directed process could.

"Electronic warfare moves at machine speed, and operational advantage depends on autonomy," said Christian Gutierrez, Senior Vice President of Hivemind at Shield AI. "This flight test proved the system can compress the sensor-to-decision cycle in real time, enabling autonomous systems to sense, share, and act on spectrum threats faster than ever before."

The implication extends beyond EW. If AI can manage the most time-critical military domain, radio-frequency combat, the same autonomy stack can plausibly handle other contested operations: coordinated drone swarms, air defense suppression, and multi-domain command and control. DiSCO paired with Hivemind is a reference architecture for how AI-driven spectrum battle management could integrate into the Pentagon's Joint All-Domain Command and Control (JADC2) framework.

The Technology Stack Behind the Flight

DiSCO is a software-defined electromagnetic battle management ecosystem, developed by the company over several years. It detects, collects, and analyzes known and unknown threat signals, creating a fused common operating picture of the electromagnetic environment. The system is designed as an open architecture, allowing integration with third-party autonomy software.

Shield AI's Hivemind is the AI pilot layer. Originally developed to fly fighter jets and drones autonomously, it operates as mission-agnostic autonomy: it can be layered onto any platform: aircraft, ground vehicle, or naval vessel, and interpret sensor data from any source. The startup, valued at $5.3 billion after a $240 million raise in March 2025, counts L3Harris as an investor alongside Andreessen Horowitz and Hanwha Aerospace.

The Deceptor payload is a compact, software-defined electronic warfare system deployed across multiple UAVs in the test. Its role was to sense and characterize unknown electromagnetic threats, share data back through DiSCO, and execute any electronic attack or defensive measures Hivemind directed.

The Green Wolf platform, built by the same company, is a launched-effects unmanned aircraft designed for multi-mission operations including electronic warfare, intelligence surveillance and reconnaissance, and communications relay. The flight test validated its role as a networked EW node that can carry both sensing and electronic attack capabilities simultaneously.

Industry Context: The Cognitive EW Race

The joint test is not occurring in isolation. Thales published a white paper in June 2026 on AI for electromagnetic warfare, arguing that "cognitive EW" constitutes a major force multiplier. Abaco Systems launched a SOSA-aligned single-board computer for EW workloads in March. The Pentagon's Chief Digital and AI Office awarded an $800 million agentic AI contract in late 2025, signaling that the Department views AI-driven autonomy as a core procurement category.

It is the second-largest defense tech startup in the US by valuation, behind only Anduril, and the partnership gives it a path into prime-contractor scale. It reports $21.9 billion in annual revenue with a portfolio spanning communications, spectrum dominance, and space systems, providing the integration and production capacity the startup lacks on its own.

Nexithon Forecast

Development scenarios

Sources